Payment card forensic analysis: From concepts to desktop and mobile analysis tools

Payment card forensic analysis: From concepts to desktop and mobile analysis tools

Digital Investigation 11 (2014) 143e153 Contents lists available at ScienceDirect Digital Investigation journal homepage: www.elsevier.com/locate/di...

2MB Sizes 0 Downloads 50 Views

Digital Investigation 11 (2014) 143e153

Contents lists available at ScienceDirect

Digital Investigation journal homepage: www.elsevier.com/locate/diin

Payment card forensic analysis: From concepts to desktop and mobile analysis tools ger c, R. Hormie re d T. Souvignet a, b, *, J. Hatin c, F. Maqua a, D. Tesniere c, P. Le Institut de Recherche Criminelle de la Gendarmerie Nationale (IRCGN), Digital Forensics Department (INL), 1 boulevard Th eophile Sueur, 93110, Rosny-Sous-Bois, France b PRES Sorbonne Universit es e Universit e Panth eon-Assas Paris II, 12 place de Panth eon, 75005, Paris Cedex 05, France c ENSICAEN, 6 boulevard mar echal Juin, 14050, Caen Cedex 4, France d INSA Lyon, 20 avenue Albert Einstein, 69100, Villeurbanne, France a

a r t i c l e i n f o

a b s t r a c t

Article history: Received 21 February 2014 Received in revised form 26 June 2014 Accepted 27 June 2014 Available online 16 July 2014

While one would not even consider them alike, payment cards are one of the most valuable and widely used embedded systems. Payment card systems are probably the most attacked and counterfeited. In fact, even though the use of smart cards have introduced high security capabilities, criminal activity has not been deterred and payment card fraud remains a lucrative activity. From low-tech (carding) to high-tech (man in the middle attack) fraud, all payment card based frauds require stealing or modifying card data and reusing it with a direct profit. Physical forms of fraud, such as Automated Teller Machine (ATM) withdrawals or in store payments, are mostly based on and associated with manipulated cards. Through their nefarious actions, that may include overwriting the magnetic strip data or injecting attacks on the embedded microcontroller, criminals are able to realise significant monetary gains. To effectively deal with these fraud cases, investigators have to quickly determine whether a card is authentic or a counterfeit. Currently no known easy forensic tool exists that provides a quick effective and accurate response. In this article, after having conceptualised payment cards as multi-interface embedded systems, we propose simple and fast forensic analysis methods to finally provide investigators with associated desktop and mobile forensic tools. © 2014 Elsevier Ltd. All rights reserved.

Keywords: Payment card fraud Skimming Carding Forensic tool Payment card analysis

Introduction Payment cards represent the most used non-cash means of payment, surpassing wire transfers and bank cheques, due to the extra protections they afford (The UK Cards

* Corresponding author. Institut de Recherche Criminelle de la Gendarmerie Nationale (IRCGN), Digital Forensics Department (INL), 1 ophile Sueur, 93110, Rosny-Sous-Bois, France. boulevard The E-mail addresses: [email protected], [email protected] (T. Souvignet), [email protected] (J. Hatin), [email protected] (F. Maqua), [email protected] (D. Tesniere), pierre.leger@ecole. ger), [email protected] (R. Hormie re). ensicaen.fr (P. Le http://dx.doi.org/10.1016/j.diin.2014.06.006 1742-2876/© 2014 Elsevier Ltd. All rights reserved.

 Consultatif du Secteur Association, 2014; Comite Financier, 2011). The total number of payment cards issued in the EU in 2011 reached 726906710 and the value of legitimate non-cash associated transactions within the region exceeded 3000 billion euros (Europol, 2012). To support such a large volume and value, payment cards are no longer just a simple plastic card with an account number, nor a simple magnetic stripe card. Since the end of the 20th century, payment cards are smart cards, with an Integrated Circuit (IC) moulded in the card plastic. According to Eurosmart (Eurosmart, 2013), more than 1.5 billion payment smart cards will be shipped in 2014, which will make payment cards one of the most widely distributed embedded systems.

144

T. Souvignet et al. / Digital Investigation 11 (2014) 143e153

Due to the associated ease of use and fast money they represent, payment cards are attractive items for organised criminal groups. In response, Law Enforcement Agencies (LEA) have had to develop investigative and forensic analysis methods to fight payment card fraud. It is paramount that LEA continue these efforts to address the escalating threat to the global banking industry. Payment card related fraud According to Europol (Europol, 2012), payment card fraud reaches around 1.5 billion euros per year in Europe. It is also a profitable activity for organised criminal groups which develop and exploit every possible form of this crime. Essentially, all of the alleged and associated activities are based on a two step crime: first obtain the payment card data, and then use it. Even if no serious studies have been conducted to provide a formal link between each form of data theft and each form of data usage, likely due to the complexity of this task, it is commonly admitted that payment card fraud can be classified into 2 main categories:  Card Not Present (CNP) or online frauds, where data comes from payment card breaches, phishing, or malware;  physical fraud, where data originates from lost and stolen cards, skimming, shimming,1 man in the middle attacks, Automated Teller Machine (ATM) reverse engineering, etc.

Visual interface The first and most obvious interface of every card is its visual one. The visual interface is much more standardised than might be expected. The following ISO standards define this interface:  ISO 7810 defines the plastic card physical characteristics with ID1 dimensions;  ISO 7811 defines location of “identification number line” and “name and address area”, and characteristics of readable characters (International Organization for Standardization, 2002a);  ISO 7811 also defines location of magnetic stripes (International Organization for Standardization, 2002b, 2008a, 2008b);  ISO 7812-1 defines the PAN format (International Organization for Standardization, 2006a);  ISO 7816 defines location of the contacts (International Organization for Standardization, 2004). Payment card visual interface is thus easily characterised and the following elements can be found on Fig. 1: 1. 2. 3. 4. 5. 6. 7. 8.

Primary Account Number (PAN) Cardholder name Expiration date Payment authority logo Card not present payment verification code (CVV/CVC) Manufacturer serial number Cardholder signature Holograms and UV securities

Most of the complaints are due to skimming, that consists of stealing payment card details and Personal Identification Numbers (PIN) against cardholder vigilance. Stolen data is then reused to make counterfeit cards, known as carding. These cards are then used at ATMs to withdraw money or at shops to buy goods. Investigating such complaints requires the investigators to fully understand payment card mechanisms and to be able to detect if a card is genuine or counterfeit. Payment cards In order to simplify payment card understanding and integrity analysis, we propose to consider payment cards as multi-interface embedded systems that contain both static and dynamic data. Nowadays payment card characteristics, mostly based on ISO 7813 (International Organization for Standardization, 2006b) and EMV standards (EMV book 1, 2011; EMV book 2, 2011; EMV book 3, 2011; EMV book 4, 2011; EMV book D, 2013), can be defined by 4 interfaces:    

visual interface; magnetic stripe interface; IC contact interface; IC contactless interface.

1

Skimming applied to chip-terminal transactions.

Fig. 1. Payment card visual interface.

T. Souvignet et al. / Digital Investigation 11 (2014) 143e153

Magnetic interface The second interface which is well established is the magnetic one. The magnetic interface consists of three magnetic tracks (Fig. 2). The first two tracks, called track 1 and track 2, are the most commonly associated with payment. Their location and the encoding principle are defined in the ISO 7811 standard (International Organization for Standardization, 2002b). The ISO 7813 (International Organization for Standardization, 2006b) standard describes the data written on payment card tracks 1 and 2. The main difference between the two tracks is the encoding density. The bits are slightly shorter on track 1, allowing for more information to be written (Fig. 3). On track 1 we find an alpha numeric encoding of the PAN, the cardholder name, the expiration date and the service code. The service code indicates if the card has a chip or not. Track 2 contains the same information except that the cardholder name that is not present. A one character Longitudinal Redundancy Checksum (LRC) is calculated to ensure data integrity for each track separately. Chip contact interface The third interface, chip contact (Fig. 4), is already in widespread use throughout Europe and Canada, but other regions have been slow to adopt this more secure method due to prohibitive costs. Due to recent security breaches in their magnetic stripe only payment system, the United States of America will commence the global roll-out to “Chip and PIN” payment cards (Levick, 2014). The application protocol used by payment chips is defined by EMV, acronym for Europay, Mastercard and Visa. It defines a global standard for the interoperability of the chip card with the Point Of Sale (POS) and ATM. The EMV standard describes the physical, electrical and data link layer. It is based on the ISO 7816 standard for the contact interface. The ISO 7816 standard is a command response protocol. The chip card is passive and only answers to the POS and ATM commands. In order to communicate with a chip card, the EMV protocol defines Application Protocol Data Units (APDU). These APDU can be sent directly after the initialisation sequence. The commands are called C-APDU and the responses are called R-APDU. After a command has been sent, a R-APDU is replied. This response is comprised of status words and optional data, with the status word dependent upon the

Fig. 2. Payment card magnetic stripes revealed.

145

Fig. 3. Data encoded on tracks 1 and 2.

card state. The EMV protocol requires that the status word is equal to NORMAL (0x9000) when a transaction is completed successfully. The data is encoded using the Basic Encoding Rules (BER) Tag Length Value (TLV) Abstract Syntax Notation One (ASN.1) format. According to the EMV flowchart (EMV book 3, 2011), the first two steps of a chip card transaction are the Initiate Application followed by the Read Application Data. The Initiate Application step is done through the SELECT APDU. A chip card could have multiple payment applications installed. For example, most French cards host a domestic payment application in addition to the international Mastercard or Visa payment applications. The EMV standards define 4 mandatory data objects:

Tag (hexadecimal)

Value

Description

5F24

Application Expiration Date Application Primary Account number

Expiration date of the payment application Number that identifies the account lean against the payment application Data required by the card to seal the transaction with a MAC calculation Data required by the card to seal the transaction with a MAC calculation

5A

8C

Card Risk Management Data Object List 1

8D

Card Risk Management Data Object List 2

Other optional or conditional data could also be present for cryptographic processing and personalisation. A common data object is the Cardholder Name (tag “5F20”). Another common piece of data is the Track 2 Equivalent Data (tag “57”) which is an emulation of the magnetic stripe track 2. This data is stored in records on the card that are personalised during the creation of the card before being issued. The records store the data of a payment application and can be read using the READ RECORD APDU.

Fig. 4. Payment card IC contact interface.

146

T. Souvignet et al. / Digital Investigation 11 (2014) 143e153

Chip contactless interface The final interface associated with payment cards is the recently introduced contactless interface (Fig. 5). In its 2014 contactless smartcart forecast, Eurosmart estimated that more than half of the card shipments will be generated by the banking industry (Eurosmart, 2013). The contactless interface is defined by the EMV contactless (EMV book D, 2013) specification. Some differences exist between the contact and the contactless interfaces. The first one is the communication protocol on which the EMV standard relies, since the contactless interface is based upon the ISO 14443 (EMV book D, 2013). The EMV book D defines an operating distance up to 4 cm high. This implies a transaction with the contactless card could be executed up to 4 cm from the POS payment terminal. The data present on the contactless interface can also be altered in order to ensure privacy. As an example the Cardholder Name (tag “5F20”) is often not present in the contactless interface (Mastercard, 2011), with it replaced by space (ASCII code 0x20) instead. Payment card analysis methods We have already presented the concept of multiinterface embedded systems when we introduced payment card internals. This simple concept is important when considering the development of methods to analyse payment cards. Published research in several papers discuss skimmer forensic analysis (Masters and Turner, 2007; Guo and Jin, 2010; Souvignet and Frinken, 2013) but does not focus on payment card analysis. In this paper, we propose 3 analysis steps that provide coverage for most types of fraud and meet investigative requirements. The first analysis consists of detecting if a card is genuine or counterfeit. The second deals with detection and identification of particular types of fraud. The last process consists of extracting valuable data for the investigators.

Thus, analysing a classic EMV payment card consists of:  transcribing PAN, cardholder name and expiry date from the visual interface;  extracting data from tracks 1 and/or 2 using a magnetic stripe reader;  extracting EMV data from contact chip;  extracting EMV data from contactless chip;  comparing relevant data to one another. Data common to all interfaces is comprised of cardholder name, PAN and expiration date. Cardholder name can be found on the front face of the card (embossed or screen printed), on magnetic stripe track 1, in EMV data object CARDHOLDER NAME (tag “5F20”) but should not be provided by the contactless interface (Mastercard, 2011). PAN and expiration date can be gathered from the front face of the card (embossed or screen printed), on magnetic stripe tracks 1 and 2, and in EMV data objects “Application Primary Account Number” (tag “5A”) and “Application Expiration Date” (tag “5F24”) provided by the contact and contactless interfaces. Usually, a simple comparison between retrieved PAN and expiration dates is enough to detect counterfeit card in most instances. The following example demonstrates the data retrieved from a card whose magnetic stripe would have been overwritten with data from another card. Particular fraud analysis Unfortunately, some counterfeit cards cannot be detected with such a basic approach. For these rare occurrences, advanced analyses have to be conducted. As some of these particular activities are yet unknown, analyses have to be designed on a case by case scenario.

Genuine detection analysis Thanks to our multi-interface concept, detecting counterfeit cards is as simple as comparing numbers. In fact, this method simply relies on reading data from all available interfaces and matching them together.

Fig. 5. Payment card IC contactless interface.

A practical example of such a fraud involves Man In the Middle (MIM) cards. This fraud, based on Cambridge University research (Murdoch, 2009; Murdoch et al., 2010),

T. Souvignet et al. / Digital Investigation 11 (2014) 143e153

relies on adding a microcontroller between the card contact and the legitimate card IC (cf. sample on Fig. 6). This microcontroller can intercept EMV commands and responses in order to manipulate the EMV data flow. As described in the Murdoch et al. paper, an interesting command to experiment with on a lost or stolen card is the VERIFY PIN command. In fact, if the “microcontroller in the middle” does not pass the C-APDU containing the VERIFY PIN command to the card IC but instead generates a RAPDU containing an answer status word NORMAL (0x9000), the card expects a non PIN verified transaction while the terminal is tricked into thinking that a PIN verified transaction is currently taking place. Detecting such a fraud can be very challenging for investigators particularly if the microcontroller insertion is done fairly well. An interface comparison analysis would only detect the fraudulent modification if the IC comes from another card, since this would easier for criminals to prepare and implant. If the IC is the original one, an advanced analysis is necessary. A simple and efficient custom analysis would then consist of sending a VERIFY PIN command outside any context of payment transaction, when the application is not even selected. If the card answers with a 0x9000 to this testing C-APDU, it can be declared that the card has been manipulated with a MIM based attack. This simple test, and custom fraud analyses in general, when provided to local digital investigators or first responders could be a great way for them to quickly determine if someone is in possession of manipulated cards.

147

Thus EMV provides an interesting feature for investigators: log entries. By “provid[ing] support for accessing a transaction log file by special devices” (EMV book 3, 2011), some EMV payment cards offer a variable size payment and withdrawal history. This feature is optional and may vary between payment schemes, issuers and even personalisation versions. For example, we have not found any American Express card with transaction log entries while most French cards disclose EMV card history (from 10 entries to dozens). If a Log Entry data element is provided (tag “9F4D”) when the application is selected, it indicates that the card supports transaction logging whose format is defined by the Log Format data element (tag “9F4F”). For example, a Log Format data element equal to “9f02069f27019f1a025f2a029a039c01” indicates that the transaction log records have the following content: Amount, Authorised (6 bytes), Cryptogram Information Data (1 byte), Terminal Country Code (2 bytes), Transaction Currency Code (2 bytes), Transaction Date (3 bytes) and Transaction Type (1 bytes). To get the log entries, stored based on a First In First Out rule, it is necessary to read the records mentioned in the Log Entry data element and then apply the Log Format to them. For example, a “000000001190400250097813080300” log entry means an accepted (40) payment (00) made in France (250) on the 3rd of August 2013 (130803), for an amount of 11,90 (000000001190) euros (0978). As dozens of records may be available, transaction logs can be a valuable source of information for investigators in order to immediately prove an individuals presence in a mentioned country.

Stored data analysis Local investigators are not all involved in investigating payment card crimes but payment cards can be a very interesting source of information. Such information is especially valuable if it can be accessed in-the-field and not after time consuming requests. Payment card stored data can provide investigators with transaction history and issuer details. Transaction log The advent of smart cards within the payment card context does not only provide security but also memory.

Fig. 6. MIM card example from Bond et al. article (Bond et al., 2014).

Resolve issuer from PAN Another typical solution for retrieving payment card history is to contact the card issuer who will be retaining all the card history; however, getting the issuer fraud department details might be quite challenging and generally requires previously engaged inside contacts or the time consuming establishment of new contacts within the payment related fraud service. Analysing the PAN can also be a method of identifying the issuing bank. The ISO 7812-1 standard breaks down the PAN into 3 different parts: the Issuer Identification Number (IIN e 1 digit), the individual account number (max 12 digits) and a check digit (1 digit) (International Organization for Standardization, 2006a). The first digit of the IIN is known as Major Industry Identifier (MII) where 3, 4 and 5 belong to the banking industry and 7 belongs to the petroleum industry. To check digit is based on a Luhn checking algorithm. Difficulties in getting the issuing bank from the IIN would require querying the American Bankers Association as stated by the American National Standards Institute, which is in charge of IIN registration (American National Standards Institute, 2014), but would only provide payment scheme details. To solve this issue, several websites provide Bank Identification Number (BIN e IIN applied to bank industry) databases or lists; however, most of them are outdated or not very accurate. In order to resolve the issuer from the PAN to obtain even more details, we contacted the main payment

148

T. Souvignet et al. / Digital Investigation 11 (2014) 143e153

schemes (AMEX, Mastercard and Visa) to gain access to their directories or, at least, relevant local focal points. By having access to these directories, investigators can resolve the issuing bank, getting their name and also obtaining fraud service addresses and phone numbers. Providing such a service to field investigators can also save precious time by aiding investigators with the ability to contact the right person. Tools

Fig. 8. Magnetic stripe data acquisition.

General introduction We have previously seen that several analyses can be conducted on the widely distributed embedded systems contained within payment cards. However, investigators efforts are inhibited by the lack of software to conduct such forensic analyses. To alleviate these shortcomings, we are also proposing two tools to meet their requirements. The first one is a desktop tool to process payment cards in order to verify card integrity, check some known discrepancies and extract/provide valuable data to the investigators, including transaction logs and issuer details. The second one is a mobile version of this tool to allow the extraction of payment card data in the field in order to conduct a first responder analysis. Both of these tools have been designed as and with open source software utilising inexpensive readers. Desktop tool Targeted systems As most police officers' terminals are workstation, we first designed a desktop tool to conduct computer assisted payment card analysis. Due to the variety of operating systems (Windows, Linux, Mac OS) used across the European police forces, the tool was developed in JAVA (1.6 and higher) which is known for its portability. Java Runtime Environment (JRE) 1.6 includes the package javax.smartcardio which was defined in the Java Portlet Specification 2.0 (JSR 268). It defines a Java API for communication with Smart Cards using ISO 7816-4 APDUs. It thereby allows Java applications to interact with Smart Cards, to store and retrieve data on the card with a PC/SC reader. Card processing Visual interface. The data acquisition is performed manually. The pictures of the front and of the back of the card can

Fig. 7. Visual data acquisition.

be imported from pictures files or captured directly with a webcam. These pictures are saved to confirm the data entries. The handwritten data will be used as reference for comparison with the data extracted from the other interfaces. When the user confirms his input, the PAN is checked according to the Luhn algorithm. If it is stated as invalid, the user is asked to verify it (Fig. 7). This is a first clue regarding the card's integrity. Magnetic interface. We chose to use a “MSR90” 3 tracks magnetic stripe Universal Serial Bus (USB) reader. This reader has been chosen due to its low cost of around 20 euros. First, this reader decodes the magnetic stripe data with Longitudinal Redundancy Checksum (LRC) and sends it to the application as keyboard strokes. Then the string is checked using a regular expression to ensure that the data is compliant with ISO 7811 part 2 specifications and contains payment data. Data from track 1 and/or track 2 are extracted from the string using the split method with standardised separator characters. Finally, the magnetic stripe's payment data are compared to the data from the visual interface entered previously. Depending on comparison results, some visual hints (tick, cross, warning, question mark) regarding the card integrity is displayed (Fig. 8). Users can also get some results which are not displayed by default by clicking the “Details” button (Fig. 9). Chip contact and contactless interface. After visual data and magnetic stripe data acquisition, users can extract data from the chip. Before starting the extraction, users have to configure the application in order to use their card reader under the Settings menu. The configuration is very easy and multiple card readers can be used, including both contact readers and contactless readers. Once done, the IC data extraction can be performed following the transaction flow detailed in Fig. 10. Interesting data (PAN, cardholder name, etc.) is then extracted from these transactions to be compared with other interface values as shown on Fig. 11. For example, reading and decoding the following record (using Application File Locator tag “94”) would offer two main pieces of information:

T. Souvignet et al. / Digital Investigation 11 (2014) 143e153

149

If a contactless interface is available, users can also analyse it using a previously configured contactless card reader, in order to get the result as shown above. As some cards do not have a contactless interface, this step can be skipped in those instances. Integrity analysis and reporting After analyses, the investigator is asked to make a decision on card integrity. He or she can declare the card as being counterfeit or genuine but can also decide not to make any judgement, rather postponing for later decision. For accountability purposes, it can be useful to know what exactly happened during the analysis; therefore,

Fig. 10. Data extraction flow on the contact interface.

Fig. 9. Details.

users can get an entire log of all commands and responses by switching to expert mode. After the investigator has made his or her decision, the case can be saved and a time stamped report can be generated. The report is a summary of all data extracted from all cards included in the case, including transaction logs, ending with the decision taken by the investigator.

150

T. Souvignet et al. / Digital Investigation 11 (2014) 143e153

Fig. 11. IC data acquisition via contact interface.

Mobile tool Targeted systems According to Gartner (Gartner, 2013), 82% of the mobile phones sold during Q3 of 2013 run on the Android operating system. In order to address this dominant platform, the mobile tool for investigators was developed for devices running Android 4.0 or newer. Android phones have been able to use the USB On-TheGo (OTG) since version 3.1 (Honeycomb). USB On-The-Go introduces the concept that a device can perform both the master and slave roles. In our application we could then use a USB card reader to perform the data acquisition. More recent Android phones also propose a Near Field Communication (NFC) management that could be used to retrieve information on the contactless interface. The NFC technology has been available since version 2.3.3 (Gingerbread MR1). The mobile tool process was designed to be is very similar to that of the desktop tool. Some concessions had to be made due to limitations of the mobile platform. However, as smartphones have become more and more powerful, the ability to offer an easy to use solution for the investigators in the field has been realised. Card processing Visual interface. The data acquisition is performed manually. Pictures of the front and back sides of the card are taken. To perform these captures, the Android smart phone camera is used. Capturing these pictures provides good data integrity to ensure that no data has been mistyped. The data captured is set as a reference (Fig. 12). All applicable additional data captured on other interfaces will be compared to this. Magnetic interface. In order to extract the magnetic stripe data, a 1 euro audio jack magnetic stripe was chosen (Fig. 13). These dongles are recognised as a microphone by Android phones, giving them the ability to read the signal and transmit it as waveforms to the audio processing unit of the phone. The provided “square API” was used to decode the magnetic stripe. In order to be sure of the data acquisition, we verified both the longitudinal redundancy checksum and the format. The audio jack reader is limited to track 2 and thus cannot read track 1 data. As cardholder name is not present

Fig. 12. Visual interface data acquisition.

on track 2, we can only extract the PAN and the expiration date. Extracted data appears as a string which is checked using a regular expression to ensure that the data is compliant with the specifications of ISO 7811 part 2. We also verify that the resulting string contains payment data. Due to the lack of track 1 analysis, the cardholder name is not verified; however, PAN and expiration date are sufficient to get relevant results (Fig. 14). Smartcard contact interface. This acquisition is done through a contact smart card reader. The reader is connected using the USB On-The-Go technology as shown in Fig. 15. An API developed by the Spanish National Institute of

Fig. 13. Visual interface data acquisition.

T. Souvignet et al. / Digital Investigation 11 (2014) 143e153

151

The first one is to use the basic EMV process. After issuing a Get Processing Option command, the card normally answers with an object, the Application File Locator (AFL e tag “94”), that indicates the list of its records. We use this list to read all of the records in the card. In some case, if the data inserted in the GET PROCESSING OPTION was not suitable for the card, we were unable to recover the AFL from the card. A second method consists of retrieving data trying all possible records on the card.

Fig. 14. Magnetic interface data acquisition.

Communication Technologies (INTECO), called DNIe Droid 1.0 API (INTECO, 2012), has been used to implement a software driver that handles the card reader. To the authors' knowledge, this library is compatible with all PC/SC/ CCID readers. APDU are sent to the smart card. The first step is the selection process. The selection has been implemented following the EMV standard. On the contact interface, the PSE (1PAY.SYS.DDF01) application is usually present. This application is a directory of all the payment applications available on the card. If this application is not present, the EMV standard asks the terminal to select all known applications. We implement the same behaviour based on Java EMV Reader list (sasc, 2012). Then, we select the main application. If this application contains transaction logs (tag “9F4D”), the last transactions performed with this card can be retrieved by issuing the READ RECORDS command. Two different methods to retrieve the payment data from the records were implemented.

Fig. 15. Contact interface data acquisition.

Smartcard contactless interface. This acquisition can be done with a contactless PC/SC reader. In this case, we used the USB OTG technology as described in the contact data extraction. A second solution is to use the NFC capability of the Android device. The data extraction process is very similar to the contact interface (Fig. 16). The only difference is in the application selection process. The application directory is not the PSE (1PAY.SYS.DDF01) but rather the PPSE (2PAY.SYS.DDF01). The information is displayed differently but is the same as what is present in the contact interface. In each case, the data acquisition follows the same principles as those described in Fig. 10. Integrity analysis and reporting After each data acquisition, data integrity is verified. This is done using a simple comparison between the payment information from each interface. Depending on comparison results, some visual hints (tick, cross, warning, question mark) regarding the card integrity are displayed. For example, the following diagram (Fig. 17) shows how we choose the visual hint for the PAN. At the end of the data capture process, all the information from the card, as well as the result of this analysis, can be exported as an archive (ZIP file) that contains all the data in one XML file, including the card pictures that were taken at the first step. This archive can be transformed into a full report in PDF format with the desktop application. This does not require any additional data capture.

Fig. 16. Contactless data acquisition: choice between smart phone NFC embedded reader or OTG attached one.

152

T. Souvignet et al. / Digital Investigation 11 (2014) 143e153

solutions have been developed to enforce cardholder presence at the time of card use. Even if they dramatically reduce online fraud, strong authentication solutions based on token, password or text message, such as 3DSecure, Verified by Visa, MasterCard SecureCode, still suffer from social engineering and mobile malware vulnerabilities. Proposed solution to prevent payment card data theft An upcoming European solution to prevent skimmer/ carding issues is geoblocking which consists of blocking any payment made out of an authorised geographic area (e.g. country). Another technique, that can be seen as the first step to removing the magnetic stripe, seems to have had great results in countries such as the Netherlands and Slovenia and initial figures would be very interesting to analyse. Finally, a better solution to monitor card related crimes would be to require all stakeholders, including banks, citizens, payment schemes, terminal manufacturers, and police, work together within an alert and fraud prevention network. This would be an efficient solution to share resources usually dedicated to preventing, identifying and fighting crimes. Conclusion

Fig. 17. Data extraction flow on the contact interface.

Discussion This paper addressed card reuse fraud only by providing methods and tools for law enforcement agencies to determine payment card data status and/or retrieve valuable information from them. Payment card reuse fraud can only be possible if at first acquisition fraud already occurred. Even if is not within the scope of this current research, it is important to examine and improve solutions to prevent payment card data theft in an effort to tackle the whole fraud. Current solution to prevent payment card data theft Many solutions already exist to prevent payment card information from being stolen at every level of the payment process: issuer, cardholder, acquirer, merchant, etc. Over the last decade, the main preventative measures have come from the PCI Security Standards which issues physical and logical requirements for either proximity or online payments. These requirements include the usage of encryption and tamperproof mechanisms to prevent illegal access and usage of gathered data. However, these security countermeasures cannot prevent criminals from stealing payment card data from the cardholder himself/herself, by way of stealing cards, phishing, or using portable skimmers. Advanced payment

At approximately 1.5 billion euro per year, payment card fraud is an attractive field for organised criminal groups. Fighting this crime is a challenging issue for law enforcement agencies that have to investigate online and in the field. They also have to develop new methods to analyse and keep up with the ever-changing techniques of innovative fraudsters. Detecting counterfeit cards is one of the challenges that sometimes have to be performed by first responders. In this article we proposed some methods to extract and analyse data from payment cards. We also proposed two tools, a desktop one and a mobile alternative, to assist non specialised investigators in their duties. First results and current development Initial beta test feedback reveals that such an application is a global need for investigators who, up to now, were using some old fashioned, non-forensic, techniques, like doing a micro payment at a local store in order to get the magnetic stripe details from the merchant receipt. While beta testers were first interested by the counterfeit card detection feature, most of them also showed high interest in accessing cards' transaction logs, when available. Current developments of interest consist of integrating payment scheme directories into a law enforcement restricted version of our application. The report generated by this version would instantly communicate the fraud service details to the issuer if available. Future developments The next challenges will be to get these tools distributed and maintained. In terms of the maintenance issue, the

T. Souvignet et al. / Digital Investigation 11 (2014) 143e153

tools will be released as open source (except law enforcement restricted features) on the future Europol development platform. Such a distribution will enable large scale review and possible adoption by the majority of European police forces. As smart cards have become more and more widespread embedded systems, their analysis is increasingly valuable for the investigation. Future developments could consist of expanding these methods and tools to other card applications such as transportation, loyalty and petroleum. Ultimately, these applications have been developed with a modular and “next fraud ready” intent. With the possibility for easy EU global adoption, due to free software distribution and low cost readers, they could represent an innovative major incident response vector in the event of an emerging global fraud threat. Acknowledgements The authors would like to thank the proofreaders, especially Dan Embury from the Royal Canadian Mounted Police (RCMP) and Johan van Heerden from the South African Police Service, for their appreciated support to publish this article. References American National Standards Institute. Issuer identification number (iin) [online; last seen on 2014/02/20]; 2014. M. Bond, O. Choudary, S. J. Murdoch, S. Skorobogatov, R. Anderson, Chip and skim: cloning emv cards with the pre-play attack, 35th IEEE Symposium on Security and Privacy, 2014.  Consultatif du Secteur Financier. L'utilisation du che que en France. Comite Tech. rep; 2011. EMV book D. emv contactless specifications for payment systems, book d: contactless communication protocol; 2013. EMV book 1. emv integrated circuit card specification for payment systems, book 1: application independent icc to terminal interface requirements; 2011. EMV book 2. emv integrated circuit card specification for payment systems, book 2: security and key management; 2011. EMV book 3. emv integrated circuit card specification for payment systems, book 3: application specification; 2011.

153

EMV book 4. emv integrated circuit card specification for payment systems, book 4: cardholder, attendant and acquirer interface requirements; 2011. Europol. Payment card fraud in the European Union. Tech. rep; 2012. Eurosmart, figures; november 2013. Gartner. Gartner says smartphone sales accounted for 55 percent of overall mobile phone sales in third quarter of 2013 [online; last seen on 2014/02/08]; november 2013. URL, http://www.gartner.com/ newsroom/id/2623415. Guo H, Jin B. Forensic analysis of skimming devices for credit fraud detection. In: Information and Financial Engineering (ICIFE), 2010 2nd IEEE International Conference on, IEEE; 2010. pp. 542e6. INTECO. Dnie droid 1.0 api [online; last seen on 2014/02/08]; 2012. URL, http://zonatic.usatudni.es/dniedroid/api/. International Organization for Standardization. Identification cards, recording technique, part 1: embossing. ref. no. iso 7811:2002; 2002. International Organization for Standardization. Identification cards, recording technique, part 2: identification cards recording technique. ref. no. iso 7811:2002; 2002. International Organization for Standardization. Identification cards, identification of issuers, part 2: application and registration procedures. ref. no. iso 7816:2004; 2004. International Organization for Standardization. Identification cards, identification of issuers, part 1: Numbering system. ref. no. iso 7812: 2006; 2006. International Organization for Standardization. Indentification cards, financial transaction cards. ref. no. iso 7813:2006; 2006. International Organization for Standardization. Identification cards, recording technique, part 6: magnetic stripe e high coercity. ref. no. iso 7811:2008; 2008. International Organization for Standardization. Identification cards, recording technique, part 7: magnetic stripe e high coercity, high density. ref. no. iso 7811:2008; 2008. Levick R. Mastercard vs. target: is there a data security war ahead?; february 2014. Mastercard. Paypass e m/chip requirements; 2011. Masters G, Turner P. Forensic data recovery and examination of magnetic swipe card cloning devices. Digit Investig 2007;4:16e22. Murdoch SJ. Reliability of chip & pin evidence in banking disputes. Digital Evidence & Elec Signature L Rev 2009;6:98. Murdoch SJ, Drimer S, Anderson R, Bond M. Chip and pin is broken. In: Security and Privacy (SP), 2010 IEEE Symposium on. IEEE; 2010. pp. 433e46. sasc. Java emv reader/terminal [online; last seen on 2014/02/20]; 2012. URL, https://code.google.com/p/javaemvreader/. Souvignet T, Frinken J. Differential power analysis as a digital forensic tool. Forensic Science International 2013;230(13):127e36 {EAFS} 2012 6th European Academy of Forensic Science Conference The Hague, 20-24 August 2012, http://dx.doi.org/10.1016/j.forsciint.2013. 03.040. URL, http://www.sciencedirect.com/science/article/pii/ S0379073813001965. The UK Cards Association. Annual report 2014. Tech. rep.; 2014